Gender, Race/Ethnicity, and Unionization in Direct Care Occupations

Abstract

Objectives. The goal of this study was to measure unionization in the direct care workforce and the relationship between unionization and earnings, looking closely at differences across race/ethnicity and gender.

Methods. Using data from the Current Population Survey from 2010 to 2020, we first used logit analyses to predict the probability of unionization among direct care workers across race/ethnicity and gender. We then measured the relationship between unionization and weekly earnings.

Results. We found that male (12%) and Black (14%) direct care workers were most likely to be unionized, followed by Hispanic and other direct care workers of color. Unionized direct care workers earn wages that are about 7.8% higher than nonunionized workers, but unionized workers of color earn lower rewards for unionization compared with White direct care workers.

Conclusions. Unions are a mechanism for improving job quality in direct care work, and protecting workers’ rights to unionize and participate in collective bargaining equitably may be a way to stabilize and grow the direct care workforce. (Am J Public Health. 2022;112(11):1676–1684. https://doi.org/10.2105/AJPH.2022.307022)

Direct care workers, also known as certified nursing assistants, home health aides, and personal care assistants, face physically and emotionally challenging work as they provide basic health and personal care assistance to older adults and persons with disabilities across home, hospital, and long-term care settings. They represent a large and increasing share of the US economy, employing more than 4.6 million individuals with 1.3 million new jobs expected to be added by 2029.¹ Job quality in direct care work (e.g., wages, fringe benefits, stability, and job protection) is notoriously poor; more than one third of workers are below 150% of the federal poverty line, more than one fourth rely on government assistance (including Medicaid), and upward mobility among direct care workers is rare.^2,3 As a result, workforce shortages have long been a concern, and these became a crisis during the COVID-19 pandemic.⁴ Direct care workers have higher exits from the labor market than any other health care occupation (remaining above 6% after the pandemic).⁵ In response to poor job quality, direct care workers are taking actions like organizing and unionizing to improve conditions for themselves and their patients.⁶

It is clear that solutions are needed to improve on-the-job conditions for direct care workers and, in turn, improve recruitment and retention challenges that prohibit the health care system from functioning at full capacity. Unions, experiencing record levels of support,⁷ provide an opportunity to do so by lifting the voices of workers and centering the concerns most important to them. This is particularly important for direct care workers who are predominately women (87%) and disproportionately Black, Hispanic, or other people of color (61%) and immigrants (27%), who have historically experienced significant disadvantages in the labor market.^1,8 It is well established that unions improve job quality, particularly wages, across skill levels and industries, but past research has primarily focused on industries that are predominately White and male.^3,8,9 In this study, we aimed to understand the impact of unions and job quality within direct care work and, specifically, on the interactions of race/ethnicity, gender, and being represented by a union. Throughout this article, we use the terms “unionized workers” or “represented by a union” to indicate both workers who belong to a union and those who are covered by a union contract.

Direct care work, deeply intertwined with systemic racism and sexism, has long been devalued, meaning that wages are lower in direct care occupations compared with other occupations of that require similar skill and education levels.^10,11 These are heavily femininized occupations—disproportionately performed by women of color—and reflect the devaluation of women’s work.^12,13 Domestic workers, which include a large share of direct care workers, were originally excluded from the 1935 National Labor Relations Act along with agriculture workers to appease primarily southern Congressmembers who extracted this compromise from Franklin D. Roosevelt in exchange for not voting down the entire Act.¹⁴ While these workers are no longer excluded from collective bargaining and other rights enacted by the Act, this has been a dominant factor in suppressing wages and benefits in direct care work.

HOW UNIONS IMPROVE JOB QUALITY

Unions are recognized for improving all aspects of job quality,^17,18 most notably wages and benefits, particularly in low-skill occupations with low wages and benefits—such as direct care work.¹⁹ However, union membership, while universally low, has shifted so that middle- and high-skill workers are most likely to be unionized even though the positive union wage effects are greatest for those in low-skill occupations.^3,20 In health care, this is demonstrated by the rise of nurses’ unions, which represent around 20% of registered nurses, while around 8% of direct care and other low-skill workers are represented by a union.²¹

Commonly described as the union difference, the positive effects are largely accomplished through collective bargaining, which is the negotiation process between employers and unions (on behalf of the workers) that establishes a legally binding contract, setting wages, benefits, hours, and other conditions important to workers.^22–24 In contrast to other solutions aimed at improving job quality such as increased minimum wage or reimbursement rates, expanding access to training and education, and developing career ladders, collective bargaining offers an advantageous path for 3 distinct reasons. First, it is comprehensive of all job attributes that affect worker experiences and, by extension, their physical, mental, and financial well-being.^1,24 Second, it centers workers’ voices by focusing on the factors most important to them. Third, collective bargaining elevates worker power through unity in numbers and democratic processes.^22,24 Particularly when considering the impact for a largely marginalized workforce like direct care workers, unions—and collective bargaining—have an opportunity to amplify voices that have been historically excluded.

Importantly, the collective bargaining process is also instrumental in achieving wage equity. Unions have been effective in narrowing racial and gender wage gaps, particularly for Black and Hispanic workers and women.^23,25 As gender and race are contributing factors in the devaluation and poor job quality in direct care work, unions have the potential to mitigate the influence of racism and sexism in the existing low wages and insufficient benefits that permeate direct care work.²⁶

In this research, we aimed to understand the association between unionization and job quality in direct care occupations. Using the Current Population Survey (CPS), we first measured rates of unionization among direct care workers by gender and race/ethnicity. We then measured the association between wages and unionization, focusing on differences between men and women, and among White, Black, Hispanic, and other workers of color. Recruitment and retention of the direct care workforce is a critical public health issue, and we explored unionization as a mechanism for improving job quality in these occupations.

METHODS

We used the IPUMS CPS to analyze the relationship between wages and unionization. The CPS is a monthly US household survey conducted jointly by the US Census Bureau and the Bureau of Labor Statistics; IPUMS CPS harmonizes microdata from the monthly data from CPS.²⁷ The analytical sample included individuals that (1) were employed as a wage or salaried worker, (2) worked full time, and (3) worked in a direct care occupation, including personal care aides, home care workers, home health workers, and nursing assistants. We tested whether rates of unionization varied between institutional direct care workers and home health workers and found similar rates of unionization among both groups; for this reason, we combined both institutional and home health workers in our sample of direct care workers. The sample included 16 292 direct care workers.

Measurement

We had 2 dependent variables. In our first analysis, the dependent variable was being represented by a union. The CPS indicates whether the respondent is a union member or is a covered by a union contract in their job.

The dependent variable in our second analysis was the natural log of weekly earnings. We used a log transformation of weekly earnings to normalize the distribution of the dependent variable.²⁸ Weekly earnings were inflation-adjusted to 2020 dollars. To standardize weekly earnings across workers, we only included full-time workers in our sample and controlled for hours worked per week in our analyses.

We included a number of demographic variables in our models, including whether an individual was a woman (1) or man (0). The race/ethnicity categories we included were White (0), Black (1), Hispanic (1), and other racial/ethnic identity (1). We also included whether someone was an immigrant (1) and age and age squared. The inclusion of the squared term generates a quadratic curve, which allows the effect of age to change over the life course. We included educational attainment level as a time-varying categorical variable: high-school graduate or less (0); some college, but no degree (1); associate degree (1); or a 4-year college degree or more (1). We included 4 geographic regions in our models: the Northeast (0), South (1), Midwest (1), and West (1). We included dummy variables that indicated the calendar year of data collection (not shown in tables). Finally, we used the variable EARNWT in IPUMS USA to weight all analyses to ensure that the sample was representative of the US population.

Analyses

This article includes 2 sets of analyses. First, we used a logit regression model to predict which workers were more likely to be unionized, focusing on demographic variables as key independent variables. Second, to address our research question of the rewards for unionization, we ran a model using logged inflation-adjusted weekly earnings as the dependent variable and included unionization as the key predictor. We then calculated the predicted earnings of unionized and nonunionized workers across key demographic groups, including men and women, and racial/ethnic groups. To calculate predicted earnings of workers across key demographic groups, we ran a series of models with interaction terms between gender and unionization and race/ethnicity and unionization. These models are not included in the article but are available on request. We calculated the predicted probability of unionization and predicted earnings by using the MARGINS command in Stata (StataCorp LP, College Station, TX). All statistical analyses were conducted with Stata version 17.

RESULTS

Descriptive statistics for the sample are included in Table 1. We separated workers by whether they were represented by a union or not a union member. Workers without union coverage were the largest group, making up about 91% of workers in our sample. Among nonunion workers in our sample, the average weekly earnings were $682, 86% were women, approximately 50% were White, 22% were Black, 20% were Hispanic, and 8% identified as another race. Thirty-eight percent of nonunionized direct care workers had a high-school degree or less, followed by some college (26%), an associate degree (21%), and a 4-year degree or higher (15%).

TABLE 1—

Descriptive Statistics for Direct Care Workers: United States, IPUMS CPS, 2010–2020

	Nonunionized	Unionized
Weekly earnings, $	682.36	754.24
Observations	14 843	1 449
Population size	148 436 955	16 622 839
Demographic variables
Age, y, mean	41.7	44.3
Gender, %
Male	13.9	18.3
Female	86.1	81.7
Race/ethnicity, %
White	50.3	41.4
Black	21.6	25.6
Hispanic	20.3	21.9
Asian or another race	7.8	11.1
Born in the United States, %	79.0	71.0
Education, %
High-school degree or less	38.4	38.2
Some college	26.0	26.1
Associate degree	20.7	18.9
4-year college degree or higher	14.9	16.9
Region, %
Northeast	16.8	32.3
Midwest	19.7	16.6
South	37.2	9.4
West	26.2	41.8

Note. IPUMS CPS = Integrated Public Use Microdata Series Current Population Survey. “Unionized” indicates both workers who belong to a union and those who are covered by a union contract.

Workers who were represented by a union earned $754 per week. Approximately 82% were women, and a lower percentage of workers who were represented by a union were White (41%) compared with nonunionized workers. A higher percentage of workers who were represented by a union were Black (26%), Hispanic (22%), or Asian (11%) compared with nonunionized workers. Educational attainment was similar between unionized and nonunionized workers: 38% of direct care workers represented by a union had a high-school degree or less, followed by some college (26%), an associate degree (19%), and a 4-year college degree or higher (17%). There was significant regional variation in union representation among direct care workers; rates of union representation were highest in the West (42%) and Northeast (32%) and were lower in the Midwest (17%) and South (9%).

Likelihood of Unionization

Table 2, model 1, shows a logit model that predicts whether direct care workers were represented by a union. We found that female direct care workers were significantly less likely to be represented by a union than male workers (P < .001), but workers of color, including Black, Hispanic, and workers who identify as another race/ethnicity, were significantly more likely to be represented by a union. Educational attainment was not significantly related to union representation. Direct care workers in the Midwest, South, and West were significantly less likely to be represented by a union than workers in the Northeast (P < .05).

TABLE 2—

Models of Predictors of Unionization (Model 1) and Weekly Earnings (Model 2): United States, IPUMS CPS, 2010–2020

	Unionization (Model 1), OR (95% CI)	Logged Inflation-Adjusted Weekly Earnings (Model 2), b (95% CI)
Unionized	. . .	0.076 (0.050, 0.102)
Demographic variables
Female (Ref = male)	0.797 (0.674, 0.943)	−0.130 (−0.154, −0.106)
Race/ethnicity
White	0 (Ref)	0 (Ref)
Black	1.938 (1.636, 2.296)	−0.097 (−0.119, −0.076)
Hispanic	1.280 (1.066, 1.537)	−0.043 (−0.066, −0.020)
Asian or another race	1.322 (1.048, 1.668)	0.002 (−0.031, 0.035)
Born in the United States	1.054 (0.896, 1.240)	0.051 (0.030, 0.073)
Age	1.102 (1.052, 1.155)	0.016 (0.010, 0.022)
Age squared	0.999 (0.999, 1.000)	−0.000 (−0.000, −0.000)
Education
High school or less	0 (Ref)	0 (Ref)
Some college	1.103 (0.943, 1.291)	0.108 (0.090, 0.127)
Associate degree	1.097 (0.924, 1.304)	0.137 (0.117, 0.156)
4-year college degree or higher	1.131 (0.936, 1.366)	0.308 (0.281, 0.336)
Region
Northeast	0 (Ref)	0 (Ref)
Midwest	0.464 (0.385, 0.558)	−0.059 (−0.083, −0.035)
South	0.124 (0.099, 0.154)	−0.060 (−0.082, −0.037)
West	0.873 (0.748, 1.019)	−0.010 (−0.034, 0.014)
Other statistics
Constant	0.014 (0.005, 0.040)	6.087 (5.959, 6.215)
Survey observations, no.	67 100	66 706
R 2	. . .	0.094

Note. IPUMS CPS = Integrated Public Use Microdata Series Current Population Survey. “Unionized” indicates both workers who belong to a union and those who are covered by a union contract. Dummy variables for year were included in the model but are not shown in Table 2. The odds of being a union member were significantly higher in 2013 compared with 2010 (OR = 1.432; P < .001). Wages were significantly lower in 2012 and 2014 (b = −0.035 and −0.046, respectively; P < .05) compared with 2010, and significantly higher in 2020 (b = 0.055; P < .001) compared with 2010.

In Figure 1, we present the predicted probability of being represented by a union, calculated with the model shown in Table 2. Male direct care workers had a predicted probability of about 12% of being represented by a union, while female workers had a predicted probability of around 10% when we controlled for demographic characteristics, education, and region. Black direct care workers had the highest rate of being represented by a union, at around 14%, while Hispanic and Asian or another race direct care workers have a predicted probability of around 10%. White workers had the lowest predicted probability of approximately 8% of being represented by a union.

FIGURE 1—

Prevalence of Unionization Among Demographic Groups (Corrected by Age, Region, and Educational Level): United States, IPUMS CPS, 2010–2020

Note. IPUMS CPS = Integrated Public Use Microdata Series Current Population Survey. Whiskers indicate 95% confidence intervals. Models used for calculating predicted weekly earnings shown in Figure 1 are not shown in the article but are available on request. Models include all control models described in the Measurement section.

Unionized vs. Nonunionized Weekly Earnings

Table 2, model 2, shows a linear regression model of the natural log of inflation-adjusted weekly earnings for unionized and nonunionized direct care workers. To interpret the coefficient of the log-transformed dependent variable, we exponentiated the coefficient, subtracted 1 from this number, and multiplied by 100. We found that workers who were unionized had weekly earnings that were 7.8% higher than workers who were not unionized (P < .001) when we controlled for demographic characteristics, education, and region.

Figure 2 contains predicted weekly earnings for workers represented by a union and nonunionized direct care workers by gender and race/ethnicity. Figure 2 shows that men who were represented by a union had the higher weekly earnings ($743) compared with men who were not unionized ($690), which indicates that men who were represented by a union had weekly earnings that were 7.2% higher than those who were nonunionized, even when we controlled for demographic characteristics, education, and region. Women who were unionized earned $654 per week compared with $605 earned by nonunionized women, indicating that women who were represented by a union had weekly earnings that were 7.4% higher than those who were not unionized.

FIGURE 2—

Weekly Earnings Advantage of Unionization for Direct Care Workers by Demographic Groups (Corrected by Age, Region, and Educational Level): United States, IPUMS CPS, 2010–2020

Note. CPS = IPUMS CPS = Integrated Public Use Microdata Series Current Population Survey. Whiskers indicate 95% confidence intervals. Models used for calculating predicted weekly earnings shown in Figure 2 are not shown in the article but are available on request. Models include all control models described in the Measurement section.

White direct care workers who were represented by a union had weekly earnings of $689 compared with weekly earnings of $635 of nonunionized White workers (a difference of 7.7%), but the earnings difference between those who were represented by a union and nonunionized workers was lower for Black and Hispanic workers (4.3% and 5.7%, respectively). Workers who identified as Asian or another race had higher weekly earnings when they were unionized ($739) compared with nonunionized workers ($629), a difference of 14.9%.

Sensitivity Tests

To test the robustness of our findings, we used propensity score matching to estimate the effect of union representation on earnings while matching participants using the TEFFECTS command in Stata. The models estimated using propensity score matching strengthened the effects of union representation on wages, indicating that direct care workers who were represented by a union had wages that were 12.2% higher than their nonunion peers. Male direct care workers had earnings that were 14.0% higher, while women had earnings that were 11.4% higher when they were represented by a union. White workers had earnings that were 14.6% higher, Black workers had earnings that were 5.0% higher, Hispanic workers had earnings that were 10.9% higher, and workers who identified as another race/ethnicity had earnings that were 12.2% higher when they were represented by a union. The wage penalty experienced by Black direct care workers represented by a union was highlighted when we used propensity score matching.

To examine patterns of unionization among female direct care workers only, we included rates of unionization and predicted wages for female direct care workers only, including White, Black, Hispanic, and other workers of color, in Figures A and B (available as supplements to the online version of this article at https://ajph.org). The patterns for women only mirror our findings for all direct care workers.

DISCUSSION

Direct care jobs are a large and growing share of the health care industry and the overall US economy,²⁹ yet recruiting and retaining workers in these jobs has become a crisis issue during the pandemic.³⁰ We explored the role of unionization as a mechanism for improving wages and job quality among direct care workers. Recent research has demonstrated that unionization may have positive outcomes for patients and patient care in skilled nursing units (research on unionization in the home health context is extremely limited).^31,32 We extended this research to measure the rewards for unionization for direct care workers, with an emphasis on differences across gender and race/ethnicity.

We first examined rates of union representation among direct care workers. We found that around 11% of direct care workers were represented by a union. This is slightly lower than the national average of 12%.²¹ Consistent with national statistics of unionization, direct care workers of color were more likely to be unionized compared with their White counterparts. Black direct care workers had the highest rate of union representation, at around 14%, while Hispanic and Asian or another race direct care workers have a predicted probability of around 10%. White workers had the lowest predicted probability of around 8% of being represented by a union, which was below the national average of unionization for White workers (12%). Male direct care workers were more likely than female workers to be represented by a union.

We found that unionized direct care workers and those represented by a union earned consistently higher wages than those who were not unionized, but the rewards for unionization varied by race/ethnicity. Overall, direct care workers who were unionized had weekly earnings that were 7.6% higher than workers who were not unionized. Among male and female direct care workers, men earned higher wages, a finding that was consistent with past research on the gender wage gap among direct care workers.¹⁰ But the rewards for unionization—meaning the percent difference in wages between those represented by a union and nonrepresentation—were about the same for men (7.2%) and women (7.4%).

However, there were differences in the rewards for unionization for direct care workers of color. Black direct care workers were the most likely to have union representation, but they had the lowest rewards for union representation, with unionized Black direct care workers earning 4.3% higher wages that nonunion Black direct care workers. Hispanic workers also had lower rewards for unionization, with wages that were 5.7% higher than those of nonmembers. These findings indicate that despite workers of color organizing to gain power in the labor market, their efforts are undermined by structural racism and discrimination that devalues the work of direct care workers of color.³³

Limitations

This study had an important limitation: we were unable to track individuals over time to measure the causal impact of being represented by a union on subsequent wages. Future research should capitalize on longitudinal data that can more precisely measure the causal link between unionization among direct care workers and wage outcomes. We also did not have precise measures of location or job tenure, which are important omitted variables in predicting wages and the probability of unionization.

Public Health Implications

We explored the issue of whether unionization is an effective strategy for improving direct care occupations within the health care sector, which has a number of public health implications. First, it has become increasingly clear during the pandemic that changes need to be made in the job quality of direct care occupations to stabilize the workforce so that we can provide high-quality care for older and disabled adults who need care in the United States. The US health care system—and public health more broadly—depends on the supply of workers who have the skills needed to provide hands-on care for others, and unionization may be one mechanism for stabilizing this workforce and recruiting new direct care workers.

Second, structural racism in the labor market, linked to historical legacies of slavery and domestic service, has had a strong impact on shaping the direct care workforce; unionization has the potential to strengthen job quality and wages in these marginalized occupations, ultimately contributing to better health for this large and growing workforce and their families.

Policy Recommendations

The Biden Administration has publicly announced its support for unionizing efforts, including the Protecting the Right to Organize (PRO) Act of 2021.³⁴ Some of the PRO Act’s key features include overriding state “right-to-work” laws, which prevent unions from collecting dues from workers that they represent by contract but not membership; forbid employer interference in organizing efforts, including mandatory meetings that are often used for antiunion propaganda; permit workers to cast organizing ballots off company premises; and implement stronger penalties (financial and otherwise) to employers that violate workers’ rights.^34,35 Critically for workers in direct care jobs who are often considered self-employed or contract workers and therefore exempt from many labor laws, the PRO Act would allow them the right to unionize.³⁵ Cumulatively, this Act would provide workers in direct care jobs more protection in unionizing efforts, which could be instrumental in increasing the share of unionized workers, improving wages, and overcoming the systemic racism and sexism contributing to suppressed wages and job quality in direct care work.

Conclusions

Direct care workers are an integral part of the US health care system, particularly in providing and supporting services for older adult and disabled populations. However, job quality is poor across these occupations with low wages, few benefits, unstable hours, and limited job protections. For the health care system and care recipients, poor job quality among direct care workers creates high turnover and threatens the stability and quality of care. Unions are a mechanism for improving job quality in direct care work, and stronger supports, such as the PRO Act, are needed to improve workers’ rights to unionize and participate in collective bargaining equitably.